Table 7. Hyperparameters optimization search space for the GCNN model after phase 1 and optimized hyperparameters of phase 2 of the GCNN model development workflow.

Hyperparameters	Search space (phase 2)	Optimized hyperparameters – phase 2
Graph convolution layers	[64, 64] or [64, 64, 64]	[64, 64]
Dense layer size	64 or 128	128
Batch size	32 (fixed)	32
Dropout	X ∼ U(0.04, 0.15), step = 0.001	0.123
Learning rate	X ∼ U(0.001, 0.02), step = 0.001	0.006
Weight decay	10−5 or 10−3	10−3
Activation function	ReLU (fixed)	ReLU
Validation ROC-AUC score		0.896